[CP2K-user] [CP2K:18189] STDA parameters

[Maria Fumanal]

Dear Anna,

Many thanks for your reply and thank you for pointing out the reference.

So if I understood well the alpha and beta (or CEXP and XEXP) depend on the 
a_EX (HF fraction).
I didn't find the reference for the linear relationships that you mention, 
could you share the reference?
However, you mention it is better not to use this general linear 
relationship but to re-optimize these parameters for the particular case,
how would this be done in practice?

Many thanks in advance

Maria

El martes, 13 de diciembre de 2022 a las 13:38:03 UTC+1, hehn... at gmail.com 
escribió:

> Dear Maria,
>
> very happy to hear that you are using sTDA!
>
> MATAGA_NISHIMOTO_CEXP and MATAGA_NISHIMOTO_XEXP were added to enable more 
> flexibility in the parameter choice, with CEXP referring to the parameter 
> dubbed alpha in Eq. 28 in Ref. 
> https://pubs.acs.org/doi/full/10.1021/acs.jctc.2c00144, identical with 
> the parameter alpha of the original Grimme paper, parameterizing Coulomb 
> contributions.
> MATAGA_NISHIMOTO_XEXP analogously refers to parameter beta in Eq. 29 of 
> the reference above, also dubbed beta in the original Grimme paper, 
> parameterizing Exchange contributions.
> FRACTION is referring to a_EX in Eq. 29, thus resembling the coupling 
> parameter to define the amount of exact exchange as it is in hybrid 
> functionals.
> DO_EXCHANGE enables to switch off exchange contributions completely.
>
> Depending on the chosen ground-state reference, it is important to adjust 
> not only the amount of exact exchange, but also CEXP and XEXP. In the sTDA 
> manual of Grimme and also in corresponding papers, several values for alpha 
> and beta are suggested depending on the GS functional, differing 
> significantly from the standard choice alpha = 1.42 + a_EX *0.48 and beta = 
> 0.2 + a_EX*1.83. However, parameterization was based on molecular test sets 
> and depending on your systems, it might be beneficial to reoptimize these 
> parameters.
>
> Hope this helps, also happy to discuss further questions.
> Best regards,
>
> Anna
>
>
>
>
>  
>  
>
> Am Di., 13. Dez. 2022 um 10:58 Uhr schrieb Maria Fumanal <
> mfum... at gmail.com>:
>
>> Dear CP2K users/developers,
>>
>> I have been using CP2K to compute excited states with TDDFPT, so far 
>> using KERNEL FULL with PBE0 functional.
>>
>> I made some tests with STDA and I have some questions with respect to the 
>> default parameters. So far I used KERNEL STDA including:
>> &STDA
>> FRACTION 0.25
>> &END
>>
>> I was wondering if I should be changing/tuning any other parameter 
>> besides the TB HF fraction to get better estimates. For instance so far I 
>> used the MATAGA_NISHIMOTO default values but it is said in the manual that 
>> they are method dependent.
>>
>> Could anyone tell me a little bit more about this?
>>
>> Many thanks in advance,
>>
>> Maria 
>>
>>
>> -- 
>> You received this message because you are subscribed to the Google Groups 
>> "cp2k" group.
>> To unsubscribe from this group and stop receiving emails from it, send an 
>> email to cp2k+uns... at googlegroups.com.
>> To view this discussion on the web visit 
>> https://groups.google.com/d/msgid/cp2k/e738d454-7b48-4f53-8a5b-6d0b63afd309n%40googlegroups.com 
>> <https://groups.google.com/d/msgid/cp2k/e738d454-7b48-4f53-8a5b-6d0b63afd309n%40googlegroups.com?utm_medium=email&utm_source=footer>
>> .
>>
>

-- 
You received this message because you are subscribed to the Google Groups "cp2k" group.
To unsubscribe from this group and stop receiving emails from it, send an email to cp2k+unsubscribe at googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/cp2k/bb7c7e7e-d535-492a-a17d-5821da5f8c51n%40googlegroups.com.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20221213/06d054b3/attachment.htm>